OptProxCC: OptProxCC

In Ivis4ml/PRScc: Proximal Gradient Smoothing Sparse Convex Clustering

Description

If lambda = 0, ProxCC gives us n clusters, and if lambda -> lambda_max, ProxCC merges all samples into one cluster

Usage

OptProxCC(
  X,
  U,
  Z,
  nLambda = 10,
  Gamma,
  p,
  n,
  R,
  tol,
  maxit,
  threads,
  verbose = TRUE,
  warm = FALSE,
  metric = c("eBIC", "Silhouette")
)

Arguments

X	Data matrix to be clustered. The rows are features, and the columns are the samples
U	Initial of clustering examplers. The rows are features, and the columns are the samples. Default is NULL
Z	Prefiltered weighted by K nearest neighbouring or ANN
nLambda	A regularization parameter for cluster number within penalty term Lambda * w[k] * |U_,i - U_,j|_2
Gamma	A regularization parameter the number of nonzero features within penalty term Gamma * r[k] * |U_k,|_2
p	Number of features
n	Number of observations
R	The adaptive group lasso's weights for feature sparse penalty.
tol	tolerance for convergence
maxit	max iterations
threads	Number of Cpu threads for convex clustering used in OpenMP
verbose	print details or not
warm	Warm start scheme or not. Default FALSE
metric	Optimize scheme for hyper-parameter. eBIC, Silhouette

Ivis4ml/PRScc documentation built on June 4, 2020, 9:19 a.m.